Browse, fork, route and safety-scan skills at no cost. Paid plans add publishing, production scale, and your whole team — 70 features, priced to grow with you.
Start free — the whole skills registry, router, and safety scan, plus your first agent.
Go from skill consumer to publisher — with a team of 5 and private analytics.
Run skills in production at scale — governance, 10× the traces, 25 seats.
Unlimited scale, with an org-private registry, SSO & an SLA.
The 15 differences that actually decide a tier — the meters and unlocks where the numbers jump, skills first.
| Feature | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| Skills Registry & Router | ||||
| Active skills (create + fork) | 10 | 50 | 250 | Unlimited |
| Publish skill to public registry | — | ✓ | ✓ | ✓ |
| Skill analytics: A/B compare + org leaderboard | — | ✓ | ✓ | ✓ |
| SkillSafety | ||||
| Deep on-demand safety review / month | 10 | 100 | 1,000 | 5,000 |
| Manifests & Versioning | ||||
| Versioned agents | 1 | 3 | 10 | Unlimited |
| Regression Checks & Impact | ||||
| Regression checks / month | 50 | 200 | 1,000 | Unlimited |
| Behavioral call-replay — real mode (+ LLM judge) | — | ✓ (metered) | ✓ (metered) | ✓ (metered) |
| Observability & Traces | ||||
| Traces / month | 5,000 | 10,000 | 100,000 | Unlimited |
| Trace retention | 14 days | 30 days | 90 days | 365 days |
| Evaluation & Benchmarking | ||||
| LLM-as-judge evals / month | 100 | 1,000 | 10,000 | Unlimited |
| Benchmark runs / month (A/B verified-lift) | 10 | 100 | 1,000 | 25,000 |
| Training Data / SFT | ||||
| SFT rows / month (dataset build) | 500 | 5,000 | 50,000 | Unlimited |
| SFT hosted training runs / month | — | 2 | 5 | Unlimited |
| Team, Platform & Support | ||||
| Team members (seats) | 1 | 5 | 25 | Unlimited |
| Support | Community / docs | Priority email | Dedicated + SLA | |
| Feature | Free | Core | Pro | Enterprise |
|---|---|---|---|---|
| Active skills (create + fork) | 10 | 50 | 250 | Unlimited |
| Public registry: browse, pull, fork, use, rate + leaderboard | ✓ | ✓ | ✓ | ✓ |
| SkillScore v2 + public safety band + README lift badge | ✓ | ✓ | ✓ | ✓ |
| Smart router (semantic + effectiveness-ranked menu) | ✓ | ✓ | ✓ | ✓ |
| Router stats + menu-bloat detection & add/remove suggestions | ✓ | ✓ | ✓ | ✓ |
| Skill Rater reports / month (live-quality autorater) | 100 | 1,000 | 5,000 | Unlimited |
| Publish skill to public registry | — | ✓ | ✓ | ✓ |
| Skill analytics: A/B version compare + your-org leaderboard | — | ✓ | ✓ | ✓ |
| Weekly skill degradation digest (email) | — | ✓ | ✓ | ✓ |
| Per-model effectiveness on your own skills | — | ✓ | ✓ | ✓ |
| Per-agent skill bundle assignment | — | — | ✓ | ✓ |
| Skill audit log | — | — | ✓ | ✓ |
| Named bundle templatescoming soon | — | — | ✓ | ✓ |
| Org-private registrycoming soon | — | — | — | ✓ |
| BYO embedder for smart_routecoming soon | — | — | — | ✓ |
BILLING FAQ
What counts as a trace?
One end-to-end agent run — however many LLM calls and tool calls it contains. We meter runs, not tokens.
What happens if I go over?
Nothing breaks. Ingest pauses at the cap and resumes next cycle — or upgrade in one click. We never silently bill overages.
Can I cancel anytime?
Yes. Downgrades take effect at the end of the billing cycle, and your data stays exportable on every tier.
Is the stack really open source?
Yes — the SDK, the eval runner, and the safety gate are MIT licensed, free to fork and run on any tier. Paid plans cover hosted scale, publishing, and support.
Start free. Upgrade when it earns it.